The National Defense Authorization Act (NDAA) for fiscal year 2019 is evidence the United States is developing a more robust artificial intelligence (AI) strategy. The new law includes two sections that take stock of current AI efforts, domestically and internationally, and it devises recommendations to maintain and accelerate U.S. research and adoption. In doing so, the bill demonstrates awareness of the concerns on the horizon – namely, public-private partnerships, ethics, and competition with China.
The first AI portion, Section 238, calls for the Secretary of Defense to create a strategic plan and coordinate the department’s AI efforts. To facilitate the strategic plan, the Department will leverage research relationships with “defense and private industries, research universities, and unaffiliated, nonprofit research institutions.” The provision also indirectly references the international competition surrounding this nascent technology by seeking recommendations to “educate, recruit, and retain leading talent” and leverage advances in commercial and academic research. Finally, this section calls for the development of ethical and legal AI policies in conjunction with the technology’s application.
In totality, this section aligns nicely with the newly established Joint Artificial Intelligence Center (JAIC) with its emphasis on accelerating, scaling, and synchronizing Department of Defense AI activities through public-private partnerships while “ensuring strong commitment to military ethics and AI safety.” However, the NDAA portion has a broader research mandate, conducting a comprehensive review of previous Department initiatives in addition to planning and coordinating future actions.
A second section establishes the National Security Commission on Artificial Intelligence. With membership designated by both the legislative and executive branches, the commission will review U.S. AI advances by evaluating America’s international competitiveness in “national security, defense, public-private partnerships, and investments.” Like Section 238, this body will emphasize investment in basic and advanced research across the public-private sectors, development and retention of potential AI talent, and ethical implications. However, the commission will provide a government-wide review rather than focus on the Defense Department.
The new law demonstrates an awareness, not only of the importance of artificial intelligence, but of the reality that the United States is not alone in pursuit of its development. For example, the emphasis on public-private partnerships and leveraging academia is reminiscent of China’s “military-civil fusion,” which seeks to integrate academia, the government, and the private sector. China is also maximizing talent by funding AI programs in Chinese universities and tapping America’s graduates and research. However, China is not the sole competitor; other countries such as India, Russia, and Israel are also pursuing the technology to varying degrees.
The law’s focus on AI ethics is occurring as public-private partnerships are coming under fire. Prominent artificial intelligence researchers and companies recently signed a declaration against lethal autonomous weapons. Similarly, the controversy surrounding Google’s work on the Pentagon’s Project Maven – a program that uses artificial intelligence to interpret videos and improve drone strikes – demonstrates the need to address AI ethics. The Maven debacle also illustrates the unrelenting undercurrent of competition; Google is deepening its engagement in China and expanding its AI research in the country. However, China’s highly concerning application of facial recognition and related technologies without regard for ethics or safety leaves a crucial gap of leadership in AI governance and ethics. Therefore, the inclusion of artificial intelligence ethics and safety in the NDAA is the first step for the United States to become the global leader in AI ethics and governance.
Despite its current technological superiority in artificial intelligence, the United States faces competition from numerous other nations. The NDAA acknowledges this reality and this year’s language is an initial attempt to maintain global leadership. While there is much left to do, this legislation lays the groundwork for advancing American artificial intelligence endeavors.
Kathryn Dura is an intern for the technology and national security program at the Center for a New American Security.